sed situation

Shell Programming and Scripting
1

Hi,

I'm looking for someone who can think in sed. Basically, I need the trailing characters on every line in a file to be deleted. These characters are all in capitals, and always follow a number, but they often vary in number

For instance, on the line:

2006_10_9_p20_TALK

I'd want to remove the _TALK. And on the line

2006_10_9_p119_CRITICS

I'd want to remove the _CRITICS. I'm reasonably good with sed basics, but no guru. Any suggestions?

Best wishes
Laurel Maury

2

Not knowing the exact format of your file, this might work, or at least point you in the right direction:

echo "2006_10_9_p119_CRITICS" | sed 's/_[A-Z][A-Z]*$//g'

Result: 2006_10_9_p119

3

In this case the g is not required since there will be only one match:

echo "2006_10_9_p119_CRITICS" | sed 's/_[A-Z][A-Z]*$//'

would be sufficient, this should also work:

echo "2006_10_9_p119_CRITICS" | sed 's/[^0-9]*$//'
4

This might also help you

$ echo "2006_10_9_p119_CRITICS" | sed 's/^\(.*\)_.*$/\1/'
2006_10_9_p119
$ echo "2006_10_9_p20_TALK" | sed 's/^\(.*\)_.*$/\1/'
2006_10_9_p20
5

file txt content
2006007 20001 AR 2502_TXT
2006007 20001 AU 2502_TXT
2006007 20001 CL 2502_TXT
2006007 20001 CO 2502_TXT1
2006007 20001 ES 2502_TXT

sed 's/^\(.*\)_\([A-Z]*\)$/\1/' txt
ouput :
2006007 20001 AR 2502
2006007 20001 AU 2502
2006007 20001 CL 2502
2006007 20001 CO 2502_TXT1

above command will only delete those line which matches the pattern "_[A-Z]"
--Manish